hi
a colleague has completed a logistic regression and it has been suggested
that some of the variables used may be colinear- this has also been
referred to as multicolinearity
can someone please advise me how to examine colinnearity in logistic
regression
i am aware that in regression the variance inflation factor may be used
but not sure how to obtain this in logistic regression - the package used
is spss
if colinearity is found, is the standard practice to rerun the analysis
without one of the variables which are found to be colinear
if the variables are found to be colinear what effect would this have on
the model if the colinearity is ignored
any useful references / websites on this topic would also be helpful
regards
ann
--
Ann Fitzmaurice
Medical Statistician
Dugald Baird Centre
Aberdeen Maternity Hospital
University of Aberdeen
Tel 01224 553876
*New IMMPACT recruitment opportunities*
For more details see: http://www.abdn.ac.uk/jobs
Open Day 2003 - Tuesday 26th August
For more details see: http://www.abdn.ac.uk/openday
|